Generally, traps are software interrupts. For example, traps are special instructions that are embedded in programs that cause the program to transfer control to a trap handler routine or which are explicitly called when an exception occurs. Typically, after a trap is triggered, the trap handler routine is, executed and the control is transferred back to the program that caused the trap. Traps can also be explicitly set so as to execute a trap handler when a specific condition is met. Since traps are embedded in programs, they cannot be easily changed by external means.
In wireless devices, there is a need to set traps to gather information about events due to certain conditions on the device. Not all traps can be set a priori. Some of the traps have to be remotely set, i.e., activated and the conditions for a trap trigger should be remotely programmable. Unlike traditional systems, in wireless systems it does not suffice just to execute a trap handler locally. The result of the execution of the trap handler caused by a trap trigger must be sent to a server for purposes of diagnostics. Since wireless devices are resource constrained and bandwidth is at a premium, not every trap trigger needs to be sent to a management server.
For the various traps that can be set only a sub set of them needs to be monitored and the results sent to an external management server. Just the execution of the trap handler is not sufficient. A monitoring agent on the device should check if any of the traps have been triggered and then at an appropriate time send the results to the management server.
Since there are many traps that can be set and monitored, a monitoring agent needs a structured way of enumerating the traps that have triggered. A simple list of possible traps will be too inefficient. The traps have to be organized in a hierarchy of categories so that the management server can specify categories of interest. Within each category, specific traps can be monitored.
Hence, there is a need to efficiently categorize the traps into several classes, and then monitor traps in each class on a wireless data terminal and the according to a policy send the results of a trigger to a management server.